Thesis: Ph. D. in Medical Engineering and Medical Physics, Harvard-MIT Program in Health Sciences and Technology, 2017. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 121-160). / DNA methylation is an important epigenetic mark that is linked to the regulation of gene expression. It is a critical part of controlling cellular identity and is essential for normal development. DNA methylation is generally studied by comparing methylation levels at individual cytosines or computing region-level averages to identify differential methylation. Here we extended this classic viewpoint by capitalizing on a unique feature of next-generation sequencing, which provides the methylation status of CpGs that are located on the same sequencing read and hence originate from the same DNA molecule. When comparing methylation states of CpGs on the same read, we observed different levels of discordant methylation, defined as molecules where the methylation of neighboring cytosines are not correlated. We quantified the proportion of discordantly methylated reads (PDR) in normal and cancer samples, and found that global PDR levels were elevated in cancer, suggesting widespread epigenetic deregulation. While we have not yet established the mechanistic contribution of this feature, we find that discordant methylation is linked to higher genetic diversity, greater cell-to-cell transcriptional heterogeneity, and adverse clinical outcome in chronic lymphocytic leukemia (CLL). Our analytic approach introduces a novel perspective on utilizing epigenomic sequencing data, which we anticipate will be a valuable tool in understanding the regulation of DNA methylation and its contribution to cellular identity. / by Mark Kendell Clement. / Ph. D. in Medical Engineering and Medical Physics

Thesis (S.M.)--Harvard-MIT Division of Health Sciences and Technology, 2004. / Includes bibliographical references (leaves 27-29). / Evolutionarily conserved genomic regions (ecores) are understudied, and yet comprise a very large percentage of the Human Genome. Highly conserved human-mouse non-coding ecores, for example, are more abundant within the Human Genome than those regions, which are currently estimated to encode for proteins. Subsets of these ecores also exhibit conservation that extends across several species. These genomic regions have managed to survive millions of years of evolution despite the fact that they do not appear to directly encode for proteins. The survival of these regions compels us to investigate their potential function. Development of a computational framework for the classification and clustering of these regions may be the first step in understanding their function. The need for a standardized framework is underscored by the explosive growth in the number of publicly available, fully sequenced genomes, and the diverse set of methodologies used to generate cross-species alignments. This project describes the design and implementation of a system for the identification, classification and cataloguing of ecores across multiple species. A key feature of this system is its ability to quickly incorporate new genomes and assemblies as they become available. Additionally, this system provides investigators with a feature rich user interface, which facilitates the retrieval of ecores based on a wide range of parameters. The system returns a dynamically annotated list of evolutionarily conserved regions, which is used as input to several classification schemes, aimed at identifying families of ecores that share similar features, including depth of evolutionary conservation, position relative to known genes, sequence similarity, / (cont.) and content of transcription factor binding sites. Families of ecores have already been retrieved by the system and clustered using this feature space, and are currently awaiting biological validation. / by Sunil K. Saluja. / S.M.

Thesis (S.M.)--Harvard University--MIT Division of Health Sciences and Technology, 2005. / Includes bibliographical references (p. 89-92). / Pharmaceutical productivity, as measured by annual output of new molecular entities and new therapeutic biologics, has fallen significantly since reaching a peak in 1996. According to Food and Drug Administration (FDA) data, the number of new drug approvals (new molecular entities and new biologics) fell from 50 in 1996 to 29 in 2003 (FDA-BEP database 2004). Meanwhile, non-inflation adjusted expenditures for research and development have almost doubled (PhRMA 2004). This thesis uses time series analysis to characterize historical trends in new drug introductions. Linear modeling and ARIMA modeling are employed to show that the large increase in new drug approvals in 1996 is inconsistent with previous trends. The hypothesis that the 1996 increase in new drug approvals is the consequence of additional FDA processing capacity pursuant to the implementation of the Prescription Drug User Fee Act (PDUFA) is considered and rejected, based on an analysis of the underlying causes of the increase. Next, approval trends before and after the implementation of PDUFA are compared. Notably, the percentage of new drug applications resulting in approval has increased since the implementation of PDUFA while the number of applications reviewed per year has not changed significantly. The relationship between the success ratio and drug withdrawal rates is examined, with inconclusive results. / (cont.) Finally, seasonal trends in new drug application (NDA) submissions and approvals are described for years preceding and following PDUFA. A significant plurality of NDA approvals occur in the month of December before and after the implementation of PDUFA, while December NDA submissions increase in the post-PDUFA period. The ramifications of these observations on new review guidelines introduced in PDUFA II and PDUFA III, as well as the implications for NDA submission strategy, are discussed. / by James B. Graham. / S.M.

Thesis (S.M.)--Harvard-MIT Division of Health Sciences and Technology, 2007. / Includes bibliographical references (leaves 84-85). / Mergers and acquisitions in the Medical Device Industry are the primary mode of exit for early stage companies. The focus of this thesis is to examine factors which influence the value of these M&A transactions from the target and acquiring firm perspectives and to understand the value creation that occurs. Publicly available electronic and published data sources were used to build a database of 674 M&A transactions and 113 IPO events for deals with published deal values and terms between January 1996 and October 2006. In this work, we demonstrate that transaction deal value varies between various medical device industry sectors. Factors that were shown to significantly correlate with M&A transaction deal value included the Sales of the target company, Market Capitalization value of the acquiring company, type of regulatory approval, and whether the company had venture backing prior to acquisition. M&A transactions that involved targets that were Public companies had significantly higher deal values than those that were private. Using 3-day event window analysis, returns of acquiring companies were shown to be slightly negative and significantly less than the S&P composite index returns over the same period. / (cont.) Previous studies suggest that managers in larger firms tend to use overvalued stock and empire building behavior, resulting in overbidding or pay more for acquisitions. No significant difference in deal value was associated with financing terms between cash and stock transactions. The use of earn outs had no effect on the deal value or the acquirer stock returns. Markets that are developing will continue to have increasing deal value as firms attempt to establish market share and or acquire breakthrough technologies. We found that there were positive correlation of M&A or IPO transaction value for companies in less mature markets where companies have opportunities to leap frog other companies in market position and share. Conversely, there is negative correlation of M&A or IPO transaction value for companies in more mature markets. The results are discussed in terms of the specific factors that influence the transaction value and the degree to which target and acquiring firms benefit from M&A transactions. / by Kevin Lee Ohashi. / S.M.

Thesis (Ph. D.)--Harvard-MIT Division of Health Sciences and Technology, February 2010. / Vita. Cataloged from PDF version of thesis. / Includes bibliographical references. / In addition to cancer cells, solid tumors consist of a variety of cell types and tissues defining a complex microenvironment that influences disease progression and response to therapy. To fully characterize and probe the tumor microenvironment, new tools are needed to quantitatively assess microanatomical and physiological changes during tumor growth and treatment. Particularly important, is the metabolic microenvironment defined in tumors by hypoxia (low p02) and acidity (low pH). These parameters have been shown to influence response to radiation therapy and chemotherapy. However, very little is known about spatio-temporal changes in p02 and pH during tumor progression and therapy. By modifying the technique of intravital multiphoton microscopy (MPM) to perform phosphorescence quenching microscopy, I developed a non-invasive method to quantify oxygen tension (p02) in living tissue at high three-dimensional resolution. To probe functional changes in the metabolic microenvironment, I measured in vivo P02 during tumor growth and antiangiogenic (vascular targeted) treatment in preclinical tumor models. Nanotechnology is rapidly emerging as an important source of biocompatible tools that may shape the future of medical practice. Fluorescent semiconductor nanocrystals (NCs), also known as quantum dots, are a powerful tool for biological imaging, cellular targeting and molecular sensing. / (cont.) I adapted novel fluorescence resonance energy transfer (FRET) -based nanocrystal (NC) biosensors for use with MPM to qualitatively measure in vivo extracellular pH in tumors at high-resolution. While intravital multiphoton microscopy demonstrates utility and adaptability in the study of cancer and response to therapy, the requisite high numerical aperture and exogenous contrast agents result in a limited capacity to investigate substantial tissue volumes or probe dynamic changes repeatedly over prolonged periods. By applying optical frequency domain imaging (OFDI) as an intravital microscopic tool, the technical limitations of multiphoton microscopy can be circumvented providing unprecedented access to previously unexplored, critically important aspects of tumor biology. Using entirely intrinsic mechanisms of contrast within murine tumor models, OFDI is able to simultaneously, rapidly, and repeatedly probe the microvasculature, lymphatic vessels, and tissue microstructure and composition over large volumes. Using OFDI-based techniques, measurements of tumor angiogenesis, lymphangiogenesis, tissue viability and both vascular and cellular responses to therapy were demonstrated, thereby highlighting the potential of OFDI to facilitate the exploration of pathophysiological processes and the evaluation of treatment strategies. / by Ryan M. Lanning. / Ph.D.

Thesis: Ph. D., Harvard-MIT Program in Health Sciences and Technology, 2014. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 126-139). / Bone adapts its mass and architecture in response to its mechanical environment. Yet control of this process by mechanical cues is poorly understood, particularly for unloading. Defining the fundamental mechanoregulation of bone adaptation is critical for the better understanding and mitigation of bone loss in astronauts as well as clinical conditions such as spinal cord injury, stroke, muscular dystrophy, and bed rest. The overall goal of this work was to study skeletal adaptation to varying amounts of reduced loading to help delineate the relationship between mechanical stimuli and skeletal adaptation. We first examined the relative contribution of muscle and gravitational forces to the maintenance of skeletal health in mice, using botulinum toxin (BTX) to induce muscle paralysis and hindlimb unloading to eliminate external loading on the hindlimbs, alone and in combination. BTX led to greater bone loss than hindlimb unloading, while the combination of interventions led to the most detrimental effects overall, suggesting that both muscle and gravitational forces play a role in skeletal maintenance, with greater contributions from muscle forces. We then characterized skeletal adaptation to controlled reductions in mechanical loading of varying degrees employing a novel model that enables long-term exposure of mice to partial weightbearing (PWB). We found that declines in bone mass and architecture were linearly related to the degree of unloading. Even mice bearing 70% of their body weight exhibited significant bone loss, suggesting that the gravity of the moon (0.16 G) and Mars (0.38 G) will not be sufficient to prevent bone loss on future exploration missions. Finally, since bone remodeling is highly site-specific, we used gait analysis and inverse dynamics to determine the mechanical environment during PWB, and then developed a finite element model of the tibia to resolve the local strain-related stimulus proposed to drive changes in bone mass. We found modest correlations between cortical bone architecture at different PWB levels and strain energy density. Altogether this work provides a critical foundation and rationale for future studies that incorporate detailed quantification of the mechanical stimuli and longitudinal changes in bone architecture to further advance our understanding of the skeletal response to reduced loading. / by Rachel Eliman. / Ph. D.

Thesis (Ph. D.)--Harvard-MIT Division of Health Sciences and Technology, February 2007. / Vita. / Includes bibliographical references. / Most people have two ears, but we can hear with only one ear. The ability to use two ears can substantially improve performance in many circumstances. There are times, however, when the addition of a second ear results in poorer performance (i.e., contra-aural interference). Contra-aural interference is of interest because it is not explained by current auditory models, it has theoretical ramifications, and its understanding could lead to improvements in the quality of life of the hearing-impaired. More generally, the techniques and results can be applied to fields in which information is combined across an array of sensors (e.g., vision with two eyes and radar arrays). This thesis includes both psychophysical measurements and black-box modeling of level discrimination. Level discrimination was chosen to study contra-aural interference since it has traditionally been considered a monaural task (dependent on only a single ear) even though the loudness of a sound depends on both ears (i.e., binaural). This thesis demonstrates that the ability to discriminate small changes in the level of a low-frequency target stimulus presented at one ear can be adversely affected by a distractor stimulus presented simultaneously and contra-aurally to the target. / (cont.) The thesis focuses on conditions in which the target and distractor perceptually fuse; the dominant perception of the stimulus is a compact auditory image with a salient loudness and position and a secondary image referred to as the "time-image". Contra-aural interference was greatest when the introduction of the distractor decreased the reliability of both the perceived loudness and position of the dominant-image. Although the tasks used in this thesis are artificial, their simplicity allows for detailed computational modeling. The results are consistent with a model based on non-optimal integration of the information carried by the dominant-image and the time-image. The modeling separates the effects of internal coding noise and decision noise (criterion jitter). The techniques used to separate the internal coding noise from the criterion jitter can be applied to a broad range of psychology experiments. / by Daniel E. Shub. / Ph.D.

Thesis (Ph. D.)--Harvard-MIT Division of Health Sciences and Technology, 2011. / Cataloged from PDF version of thesis. / Includes bibliographical references (p. 199-217). / In vivo multimodal, multiplexed microscopy allows real-time observation of hematopoietic cells, their stem and progenitor cells and metastatic cancer cells in their native bone marrow (BM) environment. Multiplexing has made possible detailed studies of the BM's microarchitecture, which helps define the niche of these cells; it has nonetheless been limited by the paucity of suitable probes fluorescent in the near-infrared spectrum that is favored by tissue optics. This project attempts to address this problem by developing cellular and vascular fluorescent imaging probes comprised of semiconductor nanocrystals, or quantum dots (QDs), with tunable fluorescence between 65o-8oonm and exhibiting photostability, robust quantum yield and narrow fluorescence profiles that are critical for such applications. The synthesis of alloyed CdTexSe1 x QDs will be detailed in the thesis. Reproducibility and workability in subsequent steps are emphasized in the methods. Special attention is also paid to the difference between working with alloyed versus single semiconductor QDs, especially the need to achieve physical and spectral uniformity when composition and its gradient are also variable. The steps for creating biological probes from these QD fluorophores are also described. They include overcoating, water solubilization and functionalization for cellular uptake and vascular retention. Finally, the thesis returns to its motivation and reports novel methods, developed using NIR QD vascular imaging probes, for visualizing in vivo 3-D imaging data of the murine BM and characterizing the tissue's architecture. Measuring the Euclidean distance between BM osteoblasts and blood vessels is presented to exemplify a potential platform for describing the geographic relationships between cells, molecules and structural components in any tissue. / by Juwell Wendy Wu. / Ph.D.

Thesis (Ph.D.)--Harvard--Massachusetts Institute of Technology Division of Health Sciences and Technology, 1999. / Includes bibliographical references (leaves 115-128). / by Jagesh V. Shah. / Ph.D.

Thesis (Ph. D.)--Harvard-MIT Division of Health Sciences and Technology, 2010. / Cataloged from PDF version of thesis. / Includes bibliographical references (p. 123-132). / The anterior thalamus and hippocampus are part of an extended network of brain structures underlying cognitive functions such as episodic memory and spatial navigation. Earlier work in rodents has demonstrated that hippocampal cell ensembles re-express firing profiles associated with previously experienced spatial behavior. Such recapitulation occurs during periods of awake immobility, slow wave sleep (SWS) and rapid eye movement sleep (REM). Despite its close functional and anatomical association with the hippocampus, whether or how activity in the anterior thalamus is related to activity in the hippocampus during behavioral states characterized by hippocampal replay remains unknown. Here, we monitor and compare thalamic and hippocampal activities during epochs in which rats execute a simple alternation task on a circular maze as well as during sleep periods before and after track running. We employ a neural decoding algorithm to interpret spiking activity in terms of spatial correlates during wake and REM. We analyze multi unit activity (MUA) to characterize the organization of thalamic and hippocampal populations during SWS. Consistent with their role in spatial navigation, we show that during active locomotion ensembles of thalamic and hippocampal neurons represent the spatial behavior of the rat in a coordinated fashion. However, during periods of hippocampal awake replay their spatial representations become decoupled. During REM, we demonstrate that thalamic activity replicates broad activity patterns associated with awake behavior and that both hippocampus and anterior thalamus concurrently represent similar ambulatory states. During SWS, we establish that the activities in these two areas alternate between frames of elevated firing and periods of little or no activity. We show that there is a tendency for thalamic depolarized states to start and end ahead of hippocampal activity frames. These results may shed light on how information encoded by thalamic circuits could bias or be incorporated into hippocampal replay phenomena. / Ph.D.